Grantee Research Project Results
2000 Progress Report: Over-compliance in Point Source Water Pollution
EPA Grant Number: R827972Title: Over-compliance in Point Source Water Pollution
Investigators: Horowitz, John K.
Institution: University of Maryland - College Park
EPA Project Officer: Chung, Serena
Project Period: December 15, 1999 through December 15, 2000 (Extended to December 15, 2002)
Project Period Covered by this Report: December 15, 1999 through December 15, 2000
Project Amount: $59,316
RFA: Decision-Making and Valuation for Environmental Policy (1999) RFA Text | Recipients Lists
Research Category: Environmental Justice
Objective:
The objectives of this research project are to investigate the factors that affect the quality of the wastewater that is discharged from point sources in the United States. There appears to be widespread overcompliance with some of the relevant regulations; in other words, the effluent is cleaner than it needs to be. This research project examines the nature of this overcompliance and tries to distinguish among the possible reasons. The implications for water pollution policy are discussed.Progress Summary:
This research examines point-source water pollution in the United States. All dischargers that report to NPDES were examined, although most of the data were from wastewater treatment plants. We examined monthly average biological oxygen demand (BOD) concentrations in wastewater, on a plant-by-plant basis, over an 8-year period from 1992 to 1999.There appears to be substantial overcompliance with the relevant regulations. Each plant faces a limit (usually 30 mg/L) on the monthly average concentration of its effluent. In our data, average concentrations are far below this limit, often in the range of 6 mg/L. We call this "over-compliance." Let c be the ratio of the discharge concentration to the limit; this is the compliance ratio. Thus, in this example, c = 6/30, or 0.2. Whenever c < 1, the plant was overcomplying in that month. When c > 1, the plant is in violation. In much of the research, we looked at a plant's median compliance ratio where the median is taken over all months of data.
Previous understanding of overcompliance has been quite limited. A seminal work in this area (Harrington) used a theoretical model to explain why plants might be in compliance even when penalties for violation were low. However, his model did not allow plants to overcomply; plants could do no better than c = 1.
Randomness in U.S. Water Pollution
Many of the people we have talked to about this issue (primarily employees of the U.S. Environmental Protection Agency [EPA] and state regulators) speculated that the low levels of discharges, on average, are warranted by discharge randomness. Plants are believed to pollute below their permitted level, on average, to compensate for the possibility of an unexpectedly large discharge. Thus, the main thrust of this research has been on the role of randomness.
It is worth noting that almost no other economic studies have examined discharge randomness empirically. In Harrington's model, discharges were determinate (nonrandom). Randomness in his model entered through the probability that a pollution violation would be detected. McClelland and Horowitz, in an empirical study of pulp and paper plants, argued that because EPA was supposedly focusing on long-term violators, whereas most of the randomness is on a much shorter scale, there would be little randomness in the relevant measure of discharges.
Our investigation of the data shows that discharges exhibit considerable randomness on a month-to-month basis. This randomness was much greater than we expected. If EPA were to focus on monthly discharges, or discharges over several months, then this randomness could be important in decision-making by the permit holders.
Plants face regulations that govern both monthly average concentrations and daily maximum concentrations. Of course, there is even greater randomness from day to day than at the monthly level. We believe monthly concentration is the correct measure on which to focus for three related reasons:
- EPA guidelines specify that plants should design their wastewater treatment
systems to be in compliance with the monthly permits (that is, not the daily
permits).
- EPA does not collect data on daily maximum concentrations. This makes sense
on the basis of reason 1. Of course, this point also necessitates that we
focus on monthly rather than daily compliance).
- The daily limits are set higher than the monthly limits to accommodate randomness in BOD levels. In other words, the relationship between daily and monthly limits is designed so that a plant in compliance at the monthly level will typically be in compliance at the daily level.
Findings
This long introduction on discharge randomness is necessary to understand our research findings. Our research has focused on the role of randomness. This role is more important and more interesting than we originally thought.
We used PCS data to construct plant-level measures of discharge randomness (i.e., plant-level variances.) We have data for as many as 108 months; we included in our analysis all plants with at least 15 months of data, not necessarily consecutive. In our regressions, we weighted observations by the number of months of data used in constructing the randomness measure.
Some variability in plant discharges is not random and must be removed to measure the true random component. We used two methods for removing nonrandom components. We used a CUSUM test to identify plants whose discharges underwent a statistically significant structural change. Such structural changes could be due to capital changes at the plant or changes in plant management. This is not the kind of variability whose role we seek to understand. We then analyzed the set of 764 plants for which there was no structural change (Sample A).
We also used information about permit renewals to analyze discharge behavior within a permit cycle. This yielded a set of 628 plants (Sample B).
For each plant, we used the median compliance rate and the standard deviation to construct the predicted probability of a single month's violation, using the assumption that discharges are distributed log normal. These predicted probabilities are quite low. The median probability of violation is less than 1 percent.
To see how low this probability is, note that it means a violation is expected to occur only once every 100 months. For multiple violations, which is a more likely target of EPA, the probability is even lower. The probability of four violations in a 6-month period is less than one in a million. EPA may impose a penalty of $3 million for a 4-month violation. The expected penalty for a 4-month violation is 45 cents.
Next, we analyzed the factors that affect the probabilities of violations. For each plant, we constructed the variable y = ln(median c)/ ln(c), which is proportional to the probability of a violation. Our final report describes in greater detail why we chose this formula. It is essentially the ratio of mean discharges over the standard deviation. For some plants, the permitted concentration changes over time, but for most plants it is constant. Therefore, for most plants, changes in c are identical to changes in discharges. When the permitted concentration changes, it is desirable to include this in calculating the relevant compliance and variability measures.
We regressed y on these independent variables: (1) EPA-regulated vs. state-regulated plants; (2) manufacturing vs. municipal wastewater treatment plants; (3) size of plant, measured as its design flow; and (4) zip code level economic and demographic variables.
We found that: (1) there is no statistical difference in the behavior of EPA- and state-regulated plants; (2) manufacturing plants overcomply more than wastewater treatment plants; in other words, manufacturing plants have a lower predicted probability of violation; (3) larger plants overcomply more; and (4) community-level variables such as income or minority composition appear to play only a very minor role. All of these results are conditional on the plant being in Sample A or B. Samples A and B give essentially the same results.
To further understand the role of randomness, we also examined whether plants with higher discharge variability chose lower average discharges. This is the basic implication of the randomness argument. We regressed median compliance against the standard deviation and found a significant negative relationship, as the theory implies. This relationship has not been tested or reported before to our knowledge.
This research was conducted with Sushenjit Bandyopadhyay, a graduate student in the Economics Department at the University of Maryland. In our final report, we will discuss these issues, motivations, and arguments in greater detail.
Policy Implications. This role for discharge randomness seems to imply that making water pollution discharges tradeable (i.e., tradeable permits) would allow plants to hedge their randomness and therefore to increase average discharges. Of course, this claim must be treated with caution. It is merely a preliminary hypothesis. Because we find that plants are overcomplying, and not merely appearing to overcomply due to randomness, it is necessary to understand their behavior to understand how plants would behave if discharges were tradeable.
We find extremely low probabilities of violation. We believe-although this remains only speculation at this point-that many plant managers may not understand the implications of the magnitude of their discharge variability. Thus, they may not realize how much they are overcomplying.
Future Activities:
The main body of the research has been completed. There remain many interesting issues to be pursued. However, the goals of the original research project have largely been met. We will prepare and submit the final report for this grant.Journal Articles:
No journal articles submitted with this report: View all 8 publications for this projectSupplemental Keywords:
water, chemicals, toxics, effluent, discharge, public policy, cost benefit, business, industry, clean technologies, environmentally conscious manufacturing., RFA, Economic, Social, & Behavioral Science Research Program, decision-making, Economics & Decision Making, emission levels, decision analysis, economic benefits, Clean Water Act, cost benefit, economic incentives, environmental values, non-regulatory benefits, cost/benefit analysis, environmental policy, effluent, compliance costs, legal and policy choices, public policy, regulations, cost effectiveness, over-compliance, point source waterProgress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.